[人人能懂] AI的“内心戏”：它在想什么，我们知道吗？

00:01:45 你的走法，决定了你是谁00:05:57 AI翻译哪家强？我们终于有了个“懂行”的裁判00:10:03 成长的秘密：如何让机器像高手一样思考和探索00:14:13 AI的“口是心非”：我们如何看穿它？00:18:11 让 AI 既能读书破万卷，又能下笔如有神今天介绍的五篇论文：[LG] Optimizers Qualitatively Alter Solutions And We Should Leverage This  [Google DeepMind]  https://arxiv.org/abs/2507.12224 ---[CL] TransEvalnia: Reasoning-based Evaluation and Ranking of Translations  [Sakana.ai]  https://arxiv.org/abs/2507.12724 ---[LG] Spectral Bellman Method: Unifying Representation and Exploration in RL  [Technion & Georgia Institute of Technology]  https://arxiv.org/abs/2507.13181 ---[CL] LLMs Encode Harmfulness and Refusal Separately  [Northeastern University & Stanford University]  https://arxiv.org/abs/2507.11878 ---[CL] Towards Agentic RAG with Deep Reasoning: A Survey of RAG-Reasoning Systems in LLMs  https://arxiv.org/abs/2507.09477 在小宇宙查看该单集文稿